MR exhibited impaired recognition for familiar musical instruments and impaired perceptual matching for the melodic components of novel tunes. Combined with his established deficits in recognising song lyrics, familiar tunes, and rhythmic processing (Beckwith, 2003), MR exhibits each of the agnosias in the musical branch of Peretz's taxonomy, with the exception of metricality. The selective sparing of MR's metricality lends support to Peretz's notion of distinct subsystems for the temporal and melodic organisation of tunes, as well as separate modules for rhythmic and metric processing within the temporal dimension (Peretz & Coltheart, 2003).
Within melodic organisation, MR's contour representation was impaired, supporting Peretz's claim that the right STG is necessary for this process (Liegeois-Chauvel et al., 1998). MR showed impaired performance for tasks that provided a "clean" assessment of contour representation function. MR's normal performance on Peretz's Contour task may reveal a task difficulty effect or a cognitive approach effect. Her task was based on multi-componential stimuli involving temporal cues and contained tones that were presented relatively slowly. Thus, MR may have mentally labelled the outcome of the first melody (up/down/same) and then compared that verbal label with the second melody. The unsubtle cues found in the Octave-dispersed/contour-preserved task may have allowed MR to use a similar strategy. This strategy appears to have been thwarted when more demands were placed on the cognitive system, as shown by MR's inability to discriminate linearly-transformed/contour-preserved melodies.
MR's combination of impaired contour and interval processing supports Peretz's theory that contour representations provide the anchorage points for encoding interval information (Peretz, 1990). Based on the frequent co-occurrence of impaired tonality and intervallic processing, Peretz has further claimed that intervallic processing is required for the perception of tonality since the musical scale is dependent on precise intervals (Peretz, 1993b, 2001a). MR's case provides additional evidence to her claim.
Although the lowest levels of Peretz's taxonomy, melodic and temporal organisation, explain MR's music agnosia,
they do not account for his musical illusions or his incorrect perception of identical melodies.
Due to the confounding influence of an apparent perceptual deficit, MR's findings challenge Peretz's
assumption that melodic impairments arise solely as a result of an impaired and isolable contour/interval/tonality subsystem
(Peretz & Babai, 1992; Peretz & Coltheart, 2003; Peretz & Kolinsky, 1993).
Instead, MR's case suggests that perceptual deficits may underlie his melodic disorders.
MR exhibited a pitch perception deficit for individual tones presented at the level of tens of milliseconds. This highly specific deficit can be interpreted in terms of a spectral and temporal processing difficulty for the discrimination of pitches with abrupt onsets. Distorted complex pitch perception of individual notes would result in an inability to form correct melodic representations, one consequence being an impaired ability to hear the similarity of identical musical sequences. MR's higher-level deficit for identical melodies can therefore be understood as a consequence of his lower-level spectral and temporal difficulty for tones of short temporal duration.
Spectral and temporal information, through their role in timbral discrimination, is important for the recognition of musical instruments (Samson & Zatorre, 1994). Each item on the instrumental identification task contained an instrument playing relatively fast melodies or scales over a range of notes. Thus, MR's impaired instrumental identification is a plausible extension of his spectro-temporal deficit for tones of short temporal durations.
MR exhibited preserved pitch perception for patterns of tones at the level of hundreds of milliseconds or greater, necessary for processing slower sequences of segmented sound (Griffiths, 2001). Thus, MR's preserved temporal and spectral processing for tones of longer durations explains his ability to determine contour direction for Peretz's task, which presented tones at the level of hundreds of milliseconds. Unlike Peretz's cognitive module approach, a perceptual account substantiates the possibility of MR's strategy-use on Peretz's task as well as his inability to use the same strategy for faster contour tasks.
MR's preserved pitch pattern perception at the level of higher-order structure provides a possible explanation for his preserved metricality despite a rhythmic processing deficit (Beckwith, 2003). The ability to divide patterns of pitches into a stable beat relies on intact pitch pattern perception for segmented sounds (Griffiths, 1999). However, rhythmicity requires more than the ability to keep a tightly regulated periodic beat (Patel, 2003). It relies on the ability to group sounds of varying temporal durations within a regular beat, which in turn involves accurately processing fine temporal variations. MR's impaired fine temporal processing thus provides a possible explanation for his impaired rhythmic processing.
Griffiths' model of musical illusions asserts that relatively accurate pitch perception within a broadly impoverished auditory percept can cause spontaneous and inappropriate activation of stored musical representations (Griffiths, 2000). MR's pitch perception meets this criterion for tones presented for musically-relevant durations, and his bilateral hearing loss creates the requisite auditory impoverishment. If the limited pitch perception that MR achieves is sufficient to activate existing musical memories, then Griffiths' model may account for his musical illusions. Additionally, Griffiths' model requires normal pitch perception of patterned sound (Griffiths, 2000). In accordance with this notion, MR displayed a preserved ability to perceive pitch patterns. It should be noted, however, that MR's hearing loss was only detected after his stroke. Consequently, the possible causal effect of acquired deafness is confounded with his acquired brain lesion.
According to MR's description of his musical illusions, he hears normal patterns of pitch that are familiarly pleasant in terms of sound quality, but unrecognisable in terms of an identifiable tune. Conversely, Griffiths' model predicts that musical illusions should be recognisable because they are activations of stored representations (Griffiths, 2000). Therefore, it is possible that MR has impaired recognition at an associative level, which is inhibiting interaction between the perception/imagery and recognition modules. This explanation gives rise to the possibility that MR's music recognition problems, unrelated to his illusions, may not be the sole result of an apperceptive problem. While apperceptive and associative forms of agnosia are not exclusive (Griffiths, 2002), such an explanation remains speculative at this stage and MR's inability to recognise his illusions may have other causes. For example, prior to his stroke, MR idly composed novel melodies that he sung sub-vocally to himself and the illusions may be the activation of these untitled melodies.
Griffiths' (2000) musical hallucinosis participants showed more severe, spontaneous, recognisable illusory episodes than MR,
but phenomenologically, MR's case is largely consistent with Griffiths' model. However, Griffiths' (2000) proposes that the
right STG, which was partially damaged by MR's stroke, is an important brain structure underlying musical illusions.
MR's neurological difference may be related to his expressional difference from Griffiths' participants.
MR exhibited impaired sequential streaming. A critical aspect of music function is the analysis of perceptual patterns where individual notes are combined sequentially to form melodies (Bregman, 2002; Griffiths, 2003). Without this ability, one would expect a difficulty in forming coherent musical information streams that could be analysed for contour content. Accordingly, MR's inability to form contour representations in fast sequences may not only be due to a pitch perception deficit for brief tones, but also a sequential streaming deficit. Given that MR was unable to stream pitches according to Gestalt grouping rules, his cognitive system may have lacked the necessary patterned streams from which to mentally abstract the contour of melodies. MR's lesion location supports the proposition that the STG anterior to Heschl's gyrus is instrumental in processing patterns of information from one sound source (Warren et al., 2003). In this way, an auditory scene analysis account affords lower-level insight into the potential anatomical-functional correlations of contour representation.
MR's impaired intervallic processing accords with a sequential streaming deficit. Impaired sequential streaming, despite preserved pitch perception for patterns within a longer temporal window, would prevent the cognitive system from eliciting contour representations from which to anchor intervallic information. A sequential streaming deficit also explains MR's impaired tonal processing for slow pitch sequences by virtue of the relationship between intervals and the musical scale.
MR exhibited an impaired ability to perceive simultaneous streams correctly. Griffiths (2003) purports that simultaneous formation of perceptual patterns is crucial to chord formation and harmonic structure. According to MR's subjective report, the richness of his perception of grand orchestral works has been attenuated post-stroke. A simultaneous streaming deficit may explain MR's sense of diminished sound for large musical works by way of an impaired ability to synthesise separate components of chords and harmonies. A simultaneous streaming focus also provides a plausible explanation for MR's agnosia for slow familiar orchestral music (Beckwith, 2003) despite preserved pitch pattern perception for slow sequences.
Abnormal grouping mechanisms have been proposed to underlie perceptual pattern organisation deficits (Johnsrude et al., 2000). Given the weight of evidence from MR's impaired sequential and simultaneous streaming, it is reasonable to conclude that MR is unable to reconstruct an accurate representation of sound patterns. Nevertheless, the conclusion that faulty grouping mechanisms are the sole cause does not automatically follow, given the confounding influence of his pitch perception problems. While normal functioning Gestalt grouping mechanisms are paramount to an accurate representation of reality (Bregman, 2002), correct pitch perception enables the rules of proximity, similarity, symmetry, good continuation, and common fate to be applied to the correct incoming pitches (Deutsch, 1999). MR's impaired auditory scene analysis offers a powerful and novel explanation of his higher level musical deficits. Yet the contribution of this impairment, independent of an account based on MR's pitch perception deficit, is impossible to assess in the musical domain since there is no feasible way to examine either in isolation.
Strong inferences based on MR's impaired auditory scene analysis cannot be made in relation to his deficits for slow musical sequences. As the streaming tasks contained rapidly presented tones, it is indeterminable whether MR's inability to elicit the streaming effect is due to a streaming impairment of a general nature, or to a streaming impairment specific to tone patterns of brief durations.